[Docs] Update ReadME details for FP4

[dsikka]

Summary:

• We should land this once the vLLM integration lands: [Quantization] Add compressed-tensors NVFP4 support vllm#18312

[github-actions]

👋 Hi! Thank you for contributing to llm-compressor. Please add the ready label when the PR is ready for review.

Note: This is required to complete the testing suite, please only add the label once the PR is code complete and local testing has been performed.

[gemini-code-assist]

Hello @dsikka, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

Summary of Changes

Hello! Gemini here, providing a summary of this pull request to help everyone quickly understand the changes.

This PR, titled "Update readme", focuses on updating the README.md file. Based on the changes, the primary intent is to clarify and expand the documentation regarding FP4 quantization support within LLM Compressor. Specifically, it updates the description of FP4 support to include activation quantization alongside weight-only quantization, notes that this support is currently preliminary, and adds a link to a new example demonstrating FP4 activation quantization.

Highlights

• Expanded FP4 Quantization Details: The main description of FP4 support in the README has been updated to explicitly mention support for quantizing both weights and activations to FP4. Previously, it only highlighted weight-only FP4 quantization.
• Preliminary Support Clarification: The documentation now clarifies that the FP4 quantization support is currently preliminary and that additional support, particularly for MoE models, is planned for future releases.
• New FP4 Activation Example: A new link has been added to the list of quantization examples, pointing to a specific example demonstrating FP4 activation quantization (examples/quantization_w4a4_fp4/llama3_example.py).

Changelog

• README.md
  • Updated the description of FP4 support to include activation quantization and note its preliminary status.
  • Added a link to the NVFP4 configuration in compressed-tensors with an updated commit hash.
  • Added a link to an example for FP4 activation quantization under the 'Applying quantization' section.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in issue comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist is currently in preview and may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments to provide feedback.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request updates the README.md to reflect new FP4 quantization capabilities. The changes are generally clear and informative.

However, the pull request description (SUMMARY and TEST PLAN) has not been filled out. Could you please provide these details? This information is important for reviewers to understand the full context and testing of these documentation changes, and for future reference.

Summary of Findings

• Incomplete Pull Request Description: The pull request description is missing the 'SUMMARY' and 'TEST PLAN' sections. This information is crucial for understanding the context, purpose, and verification of the changes.
• Link to Internal Example File: In README.md, an absolute URL is used to link to an example file (examples/quantization_w4a4_fp4/llama3_example.py) that resides within the same repository. It's generally better practice to use relative links for internal files to ensure they correctly point to the version of the file associated with that specific version of the README.

Merge Readiness

The changes to the README.md are mostly good, but there are a couple of points to address. Primarily, the pull request description needs to be completed with a summary and test plan. Additionally, there's a suggestion to change an absolute link to an internal example to a relative one for better maintainability.

I recommend addressing these points before merging. As an AI, I am not authorized to approve pull requests, so please ensure further review and approval from the team.